diff --git a/include/ShkyeraGrad.hpp b/include/ShkyeraGrad.hpp
index 1c3a480..45541a4 100644
--- a/include/ShkyeraGrad.hpp
+++ b/include/ShkyeraGrad.hpp
@@ -8,6 +8,7 @@
 #pragma once
 
 #include "core/Type.hpp"
+#include "core/Utils.hpp"
 #include "core/Value.hpp"
 #include "core/Vector.hpp"
 
diff --git a/include/core/Utils.hpp b/include/core/Utils.hpp
index 9d31731..aeaa34d 100644
--- a/include/core/Utils.hpp
+++ b/include/core/Utils.hpp
@@ -8,6 +8,7 @@
 #pragma once
 
 #include <algorithm>
+#include <chrono>
 #include <random>
 #include <vector>
 
@@ -52,4 +53,13 @@ std::enable_if_t<std::is_integral_v<T>, std::vector<T>> sample(T from, T to, siz
     return sampled;
 }
 
+template <typename Clock = std::chrono::high_resolution_clock> auto startTimer() { return Clock::now(); }
+
+template <typename Clock = std::chrono::high_resolution_clock>
+double stopTimer(const typename Clock::time_point &start) {
+    auto end = Clock::now();
+    auto duration = std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+    return static_cast<double>(duration.count()) / 1e6;
+}
+
 } // namespace shkyera::utils
diff --git a/include/core/Value.hpp b/include/core/Value.hpp
index 08fcbcd..ea3b0c1 100644
--- a/include/core/Value.hpp
+++ b/include/core/Value.hpp
@@ -11,10 +11,12 @@
 #include <functional>
 #include <iostream>
 #include <memory>
+#include <stack>
 #include <unordered_set>
 #include <vector>
 
 #include "Type.hpp"
+#include "Utils.hpp"
 
 namespace shkyera {
 
@@ -38,7 +40,8 @@ template <typename T> class Value : public std::enable_shared_from_this<Value<T>
     Value(T data);
 
     std::vector<ValuePtr<T>> topologicalSort();
-    std::vector<ValuePtr<T>> topologicalSort(std::vector<ValuePtr<T>> &sorted, std::unordered_set<Value<T> *> &visited);
+
+    inline static double topoSortTime = 0;
 
   public:
     friend class Optimizer<T>;
@@ -51,6 +54,8 @@ template <typename T> class Value : public std::enable_shared_from_this<Value<T>
     T getValue();
     T getGradient();
 
+    static double getTopoTime() { return topoSortTime; }
+
     ValuePtr<T> tanh();
     ValuePtr<T> relu();
     ValuePtr<T> sigmoid();
@@ -188,23 +193,37 @@ template <typename T> ValuePtr<T> Value<T>::pow(ValuePtr<T> exponent) {
 }
 
 template <typename T> std::vector<ValuePtr<T>> Value<T>::topologicalSort() {
+    auto timer = utils::startTimer();
+
     std::vector<ValuePtr<T>> sorted;
     std::unordered_set<Value<T> *> visited;
 
-    return topologicalSort(sorted, visited);
-}
-
-template <typename T>
-std::vector<ValuePtr<T>> Value<T>::topologicalSort(std::vector<ValuePtr<T>> &sorted,
-                                                   std::unordered_set<Value<T> *> &visited) {
-    if (visited.find(this) == visited.end()) {
-        visited.insert(this);
-        for (ValuePtr<T> val : _children) {
-            val->topologicalSort(sorted, visited);
+    std::stack<Value<T> *> stack;
+    stack.push(this);
+
+    while (!stack.empty()) {
+        auto cur = stack.top();
+        if (visited.find(cur) == visited.end()) {
+            bool hasUnvisitedChildren = false;
+            for (auto v : cur->_children) {
+                if (visited.find(v.get()) == visited.end()) {
+                    stack.push(v.get());
+                    hasUnvisitedChildren = true;
+                }
+            }
+
+            if (!hasUnvisitedChildren) {
+                stack.pop();
+                sorted.push_back(cur->shared_from_this());
+                visited.insert(cur);
+            }
+        } else {
+            stack.pop();
         }
-        sorted.push_back(this->shared_from_this());
     }
 
+    topoSortTime += utils::stopTimer(timer);
+
     return sorted;
 }